REVIEW 14 cited by
Bridging Offline and Online Reinforcement Learning for LLMs
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Bridging Offline and Online Reinforcement Learning for LLMs
read the original abstract
We investigate the effectiveness of reinforcement learning methods for finetuning large language models when transitioning from offline to semi-online to fully online regimes for both verifiable and non-verifiable tasks. Our experiments cover training on verifiable math as well as non-verifiable instruction following with a set of benchmark evaluations for both. Across these settings, we extensively compare online and semi-online Direct Preference Optimization and Group Reward Policy Optimization objectives, and surprisingly find similar performance and convergence between these variants, which all strongly outperform offline methods. We provide a detailed analysis of the training dynamics and hyperparameter selection strategies to achieve optimal results. Finally, we show that multi-tasking with verifiable and non-verifiable rewards jointly yields improved performance across both task types.
Forward citations
Cited by 14 Pith papers
-
Step-by-Step Optimization-like Reasoning in LLMs over Expanding Search Spaces
Introduces OPT* tasks and two training regimes (solver-guided online policy optimization with rank-based reward shaping and search-based offline RL) plus a theoretical link between search success and information extra...
-
Safety Alignment of LMs via Non-cooperative Games
Jointly training an Attacker and Defender LLM in a non-zero-sum game with pairwise preference judges produces a defender with much lower jailbreak success while preserving general utility.
-
Task-Dependent Evaluation of LLM Output Homogenization: A Taxonomy-Guided Framework
Proposes a task taxonomy for functional diversity in LLM outputs, validates it via user study, introduces targeted sampling to boost diversity only where needed, and presents evidence that the diversity-quality tradeo...
-
Autodata: An agentic data scientist to create high quality synthetic data
An agentic weak–strong Self-Instruct loop, optionally meta-optimized, produces synthetic data that trains small models better than standard CoT Self-Instruct across three domains.
-
Autodata: An agentic data scientist to create high quality synthetic data
Autodata trains meta-optimized AI agents to generate superior synthetic datasets, yielding performance gains over classical methods on CS research, legal, and math reasoning tasks.
-
Autodata: An agentic data scientist to create high quality synthetic data
Autodata introduces an agentic method with meta-optimization to create higher-quality synthetic data, yielding performance gains over standard methods on CS, legal, and math tasks.
-
Multi$^2$: Hierarchical Multi-Agent Decision-Making with LLM-Based Agents in Interactive Environments
Multi² decomposes LLM agent decision-making into high-level sub-goal generation with SFT and low-level action execution with RL, claims to outperform baselines across environments, and releases three hierarchical benc...
-
Smaller Models are Natural Explorers for Policy-Level Diversity in GRPO
Smaller models provide temporally correlated policy-level diversity that serves as structured exploration for training larger models in GRPO, yielding accuracy gains such as +8.8% on AIME 24 with reduced compute via t...
-
Beyond Correctness: Rewarding Faithful Reasoning in Retrieval-Augmented Generation
A turn-level faithfulness reward improves a Search-R1-style agent's Information-Think and Think-Answer faithfulness as judged by the same reward model used for training, while task accuracy is roughly unchanged.
-
LeVo 2: Stable and Melodious Song Generation via Hierarchical Representation Modeling and Progressive Post-Training
LeVo 2 presents a hierarchical LLM-Diffusion model with progressive post-training stages to generate full-length songs that balance semantic planning, track-specific acoustics, and musicality.
-
Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning
Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.
-
Multi$^2$: Hierarchical Multi-Agent Decision-Making with LLM-Based Agents in Interactive Environments
A hierarchical LLM agent with an SFT-trained subgoal planner and an offline-to-online RL-trained action executor improves long-horizon task performance and token efficiency across three interactive text environments.
-
OGER: A Robust Offline-Guided Exploration Reward for Hybrid Reinforcement Learning
OGER adds an auxiliary exploration reward built from offline trajectories and model entropy to hybrid RL training, yielding gains on math reasoning benchmarks and out-of-domain generalization.
-
OGER: A Robust Offline-Guided Exploration Reward for Hybrid Reinforcement Learning
Novice programmers completed more tasks with lower workload using GitHub Copilot versus a human partner, but reported significantly more positive and arousing emotions with the human teammate.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.